W14 Concurrent Session 5/4/2011 3:00 PM "Data Manufacturing: A Test Data Management Solution" Presented by: Fariba Alim-Marvasti Aetna Healthcare Brought to you by: 340 Corporate Way, Suite 300, Orange Park, FL 32073 888 268 8770 904 278 0524 sqeinfo@sqe.com www.sqe.com
Fariba Alim-Marvasti Fariba Alim-Marvasti is responsible for the Data Governance/Management teams at Aetna Life Insurance Company. She leads an innovative organization driving data manufacturing across Aetna along with delivery responsibilities for testing/quality assurance within the Informatics and Medical Management domains. Fariba is a results-oriented Senior IT Executive with more than twenty-five years of proven ability to lead and manage IT organizations, delivering cost-effective solutions, while maintaining productive customer relationships.
Data Manufacturing: A Test Data Management Solution Fariba Alim-Marvasti Senior Manager II Enterprise Testing and Quality Assurance Agenda Introduction DMT Processes Tools and Automation Implementing DMT Metrics and Measures Data Governance Best Practices 2
Introduction 3 Test Data Management at Aetna The Data Management Team (DMT) was set up at Aetna in 2005 with following objectives: Efficient On time test data support for unit, application, integration, regression, endto end and other testing requirements Provide realistic aswell as exception (bad) test data to increase solution quality and reduce dependency on production data Started as a small team and has now evolved as an Industry Leading Organization supporting a growing inventory of enterprise suite of applications i DMT team consists of both onshore and offshore resources supporting data manufacturing needs of 1000+ users across Aetna s IT & Business teams. 4
Need to Manufacture Test Data Advantages of using Manufactured Data as opposed to Production Data Data Confidentiality i and Security Test Coverage Reusability Quality Eliminate risk of exposure of sensitive and confidential production datatherebythereby preventfinancial costs andloss of reputation Simulate Production like scenarios (including yet to occur and future business scenarios) to improve coverage and to maintain data integrity across systems Increase efficiency through reuse of test data by eliminating redundant scenarios found in production data. Better Solution quality due to the early identification of Data Scenarios and integration of Test Data Management as part of the Software Development Life Cycle. 5 DMT Resourcing Model Recommended Resourcing Model Onshore Offshore One lead to track and assign Data Manufacturing activities At least one primary and one secondary resource for each Domain or Application supported. Additional Resources for scheduling of jobs or additional capacity as required (Offshore) Higher Demand areas such as core upstream applications and those with large volume data needs may need additional staffing Core skills required are Application knowledge, Data Analysis Skill, Technical knowledge (Database, Scripting, Automation) and Customer Service Skills 6
DMT Processes 7 Test Data Management Lifecycle Determine Types of Tests E.g. Unit, System Tests, etc. to be supported with Manufactured Data. Identify Applications (Member, Plan, etc.). Learn the Application Build Subject Matter Expertise (SMEs). Deliver Test Data Define metrics and processes to deliver data needs and quickly identify and resolve issues. Define Operations Process to communicate Data Requests, validate data received, Service Level Agreements etc. Define Data Request Templates Define data elements to be manufactured. Predefined templates help the application team and data team communicate data needs. Identify and Refresh Reference Data Non sensitive lookup data need not be manufactured and instead can be refreshed from production for test use. Environment Setup Execute purge / cleanup processes to clean out any residual production data and ensure the continuity of data as an asset. Status Reporting and Metrics Integrate For Test Data Manufacture to be successful it should be made a key part of an applications development process. 8
Data Manufacturing Integration into Project Lifecycle Phase Planning Analysis Design & Development Testing Implementation Deliverables Estimation Test Data Data and Strategy Request Completion Project Document Sheet Sheet Plan No Deliverables Activities - Plan and Analyze DMT Needs (High level estimates, scope of data work and determine special requirements) - Include test data setup tasks in the project plan. - Testing Team to determine the data needs based on the application s data manufacturing capabilities and add them into test strategy documents. - Users need to fill in Data Request sheet in the required template and submit them to DMT. - DMT team to validate the data request and manufacture / mask the data. - DMT is not required for this phase. 9 Reusability and Training Reusability : To achieve maximum reusability in DMT process, the following should be considered Build and maintain a data repository at an application/domain level Build user friendly utilities which will help the application teams to run queries to find existing data Build traceability of data for various scenarios so that existing data can be traced and reused Training : For DMT to become a success, training is an essential ingredient for both the DMT team and the rest of the Organization. DMT Training Organizational Training Data Request processes and DMT awareness procedures Data Request processes and Applications and domains procedures Technical training Privacy and compliance Application Enhancements guidelines Privacy and compliance guidelines 10
Tools and Automation 11 Why Automate Data Creation? Increased Accuracy Consistency in Delivery Process Improved Efficiency Improved Governance Free Critical Resources for Strategic Tasks Candidate Automation Tools Data Extraction Data Creation Reporting Data Validation Create an inventory of available test data Data investigation and mining Validate completeness and accuracy of batch processes Create test data at application level Reporting DMT data to management to identify risks and issues Validation and inspection of data to ensure that it confirms to set governance or compliance standards 12
Implementing DMT 13 DMT Implementation Steps Identify Executive Sponsorship to represent and support the DMT. DMT implementation depends upon various factors in the organization like system size, tools and capabilities, team size Identify the scope of data manufacturing Identify business processes, batch processes and data sources supporting the data workflow Identify a data manufacturing team Analyze and document data characteristics Build data inventories and data templates Develop Data Management Request/ Tracking Process Build tooling to generate input data from templates and data inventories Establish data management reporting mechanisms Build automation strategy for DMT 14
Metrics and Measures 15 Metrics and Measures Domain Level Metrics Domain level status reporting to Management. Performance Metrics Compare the expected received versus Delivered to actual received versus delivered. Weekly Status Metrics Report management the ongoing data support activities along with issues impacting data setup. Measure, Analyze and Improve Cost Benefit Metrics Reporting of the total effort of data manufacturing versus the domain level testing efforts. Data Request Efficiency Analysis Measure the efficiency of data requests at a release level across environments. Data Request Efficiency Metrics Provide overview of data request efficiency and depict the progress made on efficiency over certain period of time. 16
Data Governance 17 Data Governance / Masking Strategy Data Governance Board A Data Governance Board needs to be established to review and enforce mitigations to any instances of production data usage in the non production environment. The Data Governance process ensures risks are mitigated through data manufacturing, access control, and data dt masking as applicable. The Data Governance Board needs to reinforce a consistent data strategy and be comprised of respected key individual across the organization. A well documented Charter is required for handling exceptions as they arise. Data Masking Before using production data for exception requests, to avoid misuse, various masking techniques can be adopted as described below: Scrambling Swapping names or numbers Encryption and Masking Sensitive data can be encrypted Randomizing Replacing numeric fields with random numbers Look up Fields Substituting a value from a predefined list Partial De identification Maintaining the necessary data values, but substituting, removing, or randomizing the attributes remaining data 18
Best Practices 19 Measures of a Successful DMT Engagement Mapping the test requirement and data request enabling re use of existing data Minimal i exceptions / emergency requests Business / application team sign off on initial data manufacture requirements DMT s partnership with the application/domain team to understand data requirements Timely delivery within Service Levels for data in response to requests Application team and DMT clarity on using the Data Request Templates 20
Your Next Steps Create a value proposition Obtain initial funding Socialize data manufacture advantages with stakeholders and team Select a medium complexity pilot for Data Manufacturing Implement pilot Identify and implement lessons learnt Expand 21 Thank you For further information contact: Fariba Alim-Marvasti (Senior Manager II) Alim-MarvastiF@Aetna.com Mark L. Stiner (QA Technical Specialist) StinerM@Aetna.com t Anil Kedia (QA Senior Project Manager) KediaA1@Aetna.com Nalin A. Perera (Senior Consultant Architect) Pereran1@Aetna.com Aetna - Enterprise Testing and Quality Assurance 22